JeremyHerrman

Pen-based handhelds designed around writing seem to gravitate to larger form factors than phones.

I'm surprised nobody's mentioned the similarity in size to the Apple Newton MessagePad from the 90s:

Original MessagePad: 184 x 114 mm (7.25 x 4.5 in)

reMarkable Paper Pro Move: 195.6 x 107.8 mm (7.7 x 4.24 in)

No comparison in thickness though, with the Newton being about 3x the thickness of the reMarkable (19 mm vs 6.5 mm).

totally agree - they all need a human in the loop at this point. I'm constantly stopping gpt-5/sonnet 4 and steering. Unfortunately with grok it completely misses the plot constantly

fast but not smart. Fine for non-critical "I need this query" or "summarize this" but it's pretty much worthless for real coding work (compared to gpt-5 thinking or sonnet 4)

thanks for making this! I'd love to use the microphone key (fn + mic) to trigger Handy but even after turning off dictation it doesn't seem like the system allows that key to be used (I get a dialog prompting me to turn on dictation).

Set the monitor to color and the date to September 20th, 1989. Then boot while holding ⌘-⌥-C-I to see an image of the dev team.

https://compmuseum.org/blog/iici-easter-egg/

I know this book is a first hand account from Esslinger himself, but aside from your quoted passage, I've never seen Snow White refer to a specific color, only to the design language itself. Even the other mentions of Snow White in his book refer to the design language, not a color.

The first product to feature the Snow White design language was the Apple IIc, which featured a color known as "Fog" which is distinct from the Platinum used in Apple's products from 1986-1999. For a good side-by-side comparison, check out this image of an original Apple IIc (1984) and the Apple IIc Plus (1988): https://i0.wp.com/lowendmac.com/wp-content/uploads/iic-and-i...

Thanks, I understand how stereograms work and have quite a few of these IRL. I use cross-eyed method to quickly view them (albeit inverted depth) when shown on screen.

I've tried to show my point in these videos which show basically no difference between the two images when overlapped and crossfaded between the two. https://imgur.com/a/RMy3QA3

Here are some videos trying to show what I mean. I overlapped the two images on top and crossfaded between the two. Aside from some minor distortion I don't see any major differences normally found between stereo pairs.

https://imgur.com/a/RMy3QA3

yes understood that cross-eyed method inverts the depth. My point was that some of the image pairs are from the exact same perspective - so there is no stereo depth no matter if you're using cow-eyed or cross-eyed.

Is it just me or are some of these examples not actually stereo image pairs?

I'm just crossing my eyes to see the "negative" depth image but some like "McLean’s House" and "Lincoln visits General McClellan at Antietam" don't appear to have any depth changes between them.

wow what a find! Does the carrying case feature the standard apple logo or one with no "bite"?

If you ever decide to part with it let me know :)

favorite quote from this article:

"The tone of the draft isn't the only problem. The email I'd have written is actually shorter than the original prompt, which means I spent more time asking Gemini for help than I would have if I'd just written the draft myself. Remarkably, the Gmail team has shipped a product that perfectly captures the experience of managing an underperforming employee."

Couldn't agree more. Moving a file from one folder to another has a huge delay. Dragging files for spring loaded folders doesn't work well anymore.

As a user since System 7 it's so sad to see.

Sure, computers grew in popularity throughout the 90s but I wouldn't call them uncommon in the early 90s - they were all over the place especially in schools.

My kindergarten classroom had an Apple II in 1989. Our first grade classroom had old IBM PCs.

My parents bought our first computer in '94 and we were one of the later families to have one (these were actual middle class families).

As a 14 year old in 1998, I was the sole computer user in my family and my parents wouldn't let me have internet access. Of course, that didn't stop me and I used ad-supported free ISPs like K-mart Blue Light (yes, K-Mart had an internet service!) and Freeinet to dial in when my parents were asleep and download files from Hotline. I used ResEdit to hack the client binary and make the ad windows minimizable :)

What a thrill it was to connect to a fast server and download the blazing speed at 3KB/s.

Getting access to a new Hotline server could be involved. Some came with instructions to visit a set of web pages (each with ads), copy a specific letter from each site and put them all together to get the server password.

Once you were in the server, you never knew what you would find. Hotline is where I downloaded a lot of warez (like adobe products) but some servers had really strange and NSFW stuff - including the first snuff type images I had encountered.

At some point, my parents would get clues that I something was "wrong" with their phone lines. Call waiting didn't work with modems so incoming calls wouldn't come through, and word got back about the issues. Luckily I was able to buy a product called "Catch a Call" which sat between the phone line and your modem, and when an incoming call was detected it would ring, and I would rip the cord out of the computer to let the house phone ring.

I started going online more often while my parents were awake and I would just listen for them to hit the "talk" button on the 900MHz wireless handset and rip the phone cord out of my iMac as fast as I could so they wouldn't hear the modem noise through the phone. Sometimes when I wasn't fast enough, my mom would yell "Jer! Something's wrong with the phone!" and I would walk out, hit a few buttons to pretend I was fixing it, and hand it back (my mom would usually thank me for helping). She passed away a few weeks ago... I never told her about this deceit.

Eventually around 2000/2001 they got a cable modem and life with broadband was very sweet.

To really blow peoples' minds use both hands to tap the differences, keeping the left hand for the left image and the right hand for the right image!

From your POV the images are merged so your hands will look like they're tapping a single image, but from the audience's point of view you look like a savant with multi-attention!

typo here, l=2,m=0 is the orbital with the donut

For those of you curious about WHY these shapes look like the do (e.g. "why does l=0, m=0 have a donut in the middle of two lobes?"), this video from Münster University finally gave me an intuitive understanding of how these shapes arise.

https://youtu.be/Opufc3onVow

But they were the same Apple Design Keyboard and ADB Mouse II that shipped with all of the other mid 90s macs though right?

Is anyone using local LLMs to manage their email? This seems like it could be helpful to hook that up.

I also have a repertoire of songs whose keys I've memorized to function as anchors, then using relative pitch to gauge whatever note I'm trying to discern.

I'd like to skip this relative pitch translation and jump straight to recognizing each note directly, but I have gaps in coverage. I'm happy you brought up "Comfortably Numb" since it's also burned into my neurons - just now I thought about it for a moment, went to the keyboard and hit B exactly! Also love "Hook" and will never be able to forget that harmonica intro.

Some of my songs:

  * Vampire Weekend - "Mansard Roof" in F major
  * Radiohead - "Everything in its right place" first chord: C major
  * Queen - "Bohemian Rhapsody" piano part in Bb major

> remember how cheap Uber was in 2008?

ah yes, uber was so cheap checks notes... 2 years before it raised its $1.25m seed round in 2010...

VLMs like PaliGemma and Florence-2 support object detection and segmentation, so it's becoming more common to have YOLO like capabilities built into VLMs.

Another benefit of VLMs which support object detection is that they are open vocabulary, meaning you don't have to define the classes ahead of time. Additionally fine tuning tends to keep the previous detection capabilities instead of erasing all previous classes like fine tuning YOLO.

VLMs so far have never been good at counting objects or spatial relationships (e.g. the coffee is to the right of the microwave).

There are ways to help the VLM out - Set of Marks [0] from Microsoft being the most prominent, which uses segmentation to outline and label sections of the image before sending to the VLM.

Providing "speakable" labels to regions helps ground the visual abilities of VLMs and is why in this paper the performance is so much better when words are present in the grid for "Task 6: Counting the rows and columns of a grid"

0: https://github.com/microsoft/SoM

Late at night, I work in a corner of the bedroom because my wife likes me nearby. This helps not disturb her and my toddler while they sleep.

> outsourcing the creation of a bedtime story for your child is just bad parenting.

Yup anyone who reads a pre-written bedtime story to their kids is a bad parent /s

Even before ChatGPT, I was using davinci in the OpenAI playground to have a lot of silly fun with my daughter. Totally dumb stuff like asking it to create a menu for an alien restaurant with farts as an ingredient, but it had us rolling.

So, "What Do We Want Computers to Do?" - no matter what your answer, AI makes that question a lot more interesting and raises the ceiling of possibilities. I just wish the author (and others like them) would stop moralizing.

Yes, still using it several times a week mostly for work via mac virtual display. If I need to work late at night, I find putting myself in a daylight immersive environment helps me stay awake and avoids high contrast difference between a laptop screen and a dark room.

Great to know! Looks like they only made this change at the beginning of May. Prior to that you had to turn off chat history which wasn't worth it to me.

April 25, 2024: "To disable chat history and model training, navigate to ChatGPT > Settings > Data Controls and disable Chat history & training. While history is disabled, new conversations won’t be used to train and improve our models, and won’t appear in the history sidebar. To monitor for abuse, we will retain all conversations for 30 days before permanently deleting." https://web.archive.org/web/20240425194703/https://help.open...

May 02, 2024: "To disable model training, navigate to your profile icon on the bottom-left of the page and select Settings > Data Controls, and disable “Improve the model for everyone.“ While this is disabled, new conversations won’t be used to train our models." https://web.archive.org/web/20240502203525/https://help.open...

> Privacy protections are built in when accessing ChatGPT within Siri and Writing Tools—requests are not stored by OpenAI, and users’ IP addresses are obscured. Users can also choose to connect their ChatGPT account, which means their data preferences will apply under ChatGPT’s policies.

So does this mean that by default, a random Apple user won't have their ChatGPT requests used for OpenAI training, but a paying ChatGPT Plus customer will?

Does this also mean that if I connect my ChatGPT Plus account that my data will be used for training?

It just seems strange to have a lower bar for privacy for paying customers vs users acquired via a partnership.

(yes I'm aware that the "Temporary Chat" feature or turning off memory will prevent data being used for training)

For years I wondered why electron orbitals are shaped the way they are - for instance why does the 3d orbital have two lobes and a donut? Only after learning about spherical harmonics did it finally click from an intuitive sense.

Münster University has a great set of videos which helped me understand the how orbitals relate to harmonic modes of a sphere. Check them out:

Spherical Harmonics: https://youtu.be/Ziz7t1HHwBw

More info & videos: https://www.quantumreflections.net